Strong pinning in high temperature superconducting films C.J. van der Beek and M. Konczykowski Laboratoire des Solides Irradies, CNRS-UMR 7642 and CEA/DSM/DRECAM, Ecole Poly technique, 91128 Palaiseau Cedex, France A. Abal'oshev, I. Abal'osheva, P. Gierlowski, and S.J. Lewandowski Instytut Fisyki Polskiej Akademii Nauk, Aleja Lotnikow 32, 02-668 Warszawa, Poland M.V. Indenbom Laboratoire des Solides Irradies, CNRS-UMR 7642 and CEA/DSM/DRECAM, Ecole Polytechnique, 91128 Palaiseau, France and Institute of Solid State Physics, Russian Academy of Sciences, 142342 Chernogolovka, Russia S. Barbanera Istituto di Elettronica dello Stato Solido-Consiglio Nazionale delle Ricerche, Via Cineto Romano 42-00156, Roma, Italy (February 1, 2008) Detailed measurements of the critical current density j c of YBa2Cu307_2 layers of these materials. Due to the small coherence length £ and the large condensation energy of the cuprates, such vacancies are potentially very effec- tive pinning centers; their effect may in principle explain the higb critical currents encountered in YBa2Cu307_^ films.0'cl Vortex pinning by these dense defects of typical diameter D ~ 3A e£~ 3 is the ef- fective defect density and e < 1 is the anisotropy param- eter of the superconductor, e s» 0.14 for YBa2Cu307). The volume pinning force F p is obtained from F p = (WV^ 1 ) 1 / 2 , where the pinning strength W — n ( i{f 2 ), f p is the elementary force exerted by one defect on a vor- tex line, and the averaging is carried out over a unit cell of the vortex lattice. The correlated volume V c = R 2 C L C , where the correlation lengths L c and R c are the distances, respectively parallel and perpendicular to the magnetic induction, at which the relative displacements due to pin- ning ([«(JZ c> 0)-u(0,0)] a > 1 /a and ([u(Q, L c )-u(0, 0)] 2 ) 1 / 2 are equal to £. The very high measured values j c > 1 x 10 11 Am -2 (see below) imply, however, that R c does not, under usual experimental conditions, exceed the vortex spac- ing ao = (2o/v / 3i?) 1 / 2 J3 In other words, the small scale displacements of neighboring vortices are independent. An estimate of the longitudinal correlation length (or "Larkin length") L c under the condition R c = do, i shows that L c ~ 10 nm, much less than usual film thick- nesses d > 100 nm. Here jo = 4e /3v / 3 ( i > oC is the de- pairing current density, Sq = $Q/47r^ A^ fc is the typical vortex energy scale, $o = h/2e is the flux quantum, X a b is the penetration depth for supercurrents flowing in the CuC>2-layers, and (j,q — 4tt x 10~ 7 . Contrary to the case of weakly pinning films in the two-dimensional limit, in which j c cx (n^/ii) 1 / 2 , El the critical current density 1 1/2 : Jo 27 n d (p P )e 16££q 2/3 (2) is field- and thickness-independent. Using the explicit form (fp) 1 / 2 ~ \eo(D v /£) 2 for oxygen vacancies, where D v is the oxygen ion radius, one has 3c 3o ( 21n d D v V 256e£ 2/3 (3) Note, however, that underdoped twin-free Yl^CuaOy-a single crystals, in which the increase of the sustainable current density due to an increasing density of oxygen vacancies in the Cu02-layers as one underdopes the m; terial is consistent with a concomitant decrease of T c , show an exponential temperature dependence of the cur- rent density that is not observed in thin films. YBa2Cu30y_5 films typically form thcaugh two- dimensional nucleation and island growth,oli3 with lat- eral island sizes thatjacrease with growth temperature and film thickness. UH The significant thickness modu- lations 5d related to the presence of the islands and the associated variations of the vortex line energy as function of the position, lead to pinning of vortices in the thinnest portions of the film. The maximum pinning force will be close to that needed to drive a single vortex line out of such a trough. If this mechanism is dominant, the exper- imentally measured critical current density is expected to follow 3, TV 2we Sd _ . 3V3ir£ Sd $ D~d ' 3o- 2D ( B«i|) (4) with D the average island diameterfl In their detailed study of the relation between film roughness and the cur- rent density, Jooss et al. suggested that this is indeed the case.tij However, the magnitude of the critical current density and its increase as function of the film thickness d for films thinner than 2X a b could only be explained by these authors by invoking the lowered vortex line ten- sion in such thin films, in which the effective penetration depth is equal to 2X a b/d, and by assuming an uniden- tified supplementary bulk pinning mechanism. Further- more, the studyta restricted itself to fields (ioH a < 0.3 T and T = 5 K. The discovery of growth spirals with a central screw dislocation in sputtered or metal-organic chemical va- por deposited (MO-CVD) YBa 2 Cu30 7 -,5 films^ im- mediately lead to the suggestion that extended defects rather than microscopic point defects are the main pin- ning centers .□ The growth spirals appear in laser-ablated films only when the substrate is heated above 850 °C during deposition in an oxygen pressure exceeding 50 Pa, while below this temperature, the films are formed through island growth.E3 In the latter case, screw- and edge dislocations are to be expected only in the troughs between islands, effectively halting easy vortex, motion along these. Recent experiments by Dam et aiJlil showed that a "characteristic field" (or vortex density) beyond which the critical current density of YB^CusOy-^- films decreases, is correlated with the surface density of screw dislocations n s d- Sequential etching showed the screw dislocations to extend throughout the film thickness, al- lowing for the possibility that the vortices are pinned along their entire length on the npii-superconducting dis- location core of radius cq <§; The smallness of cq means that unlike amorphous columnar defects, to which the dislocation cores are often comparedjij pinning is more likely to be due to the core-induced variation of the mean free path in the vicinity of the dislocation (5n- mechanism). Supposing that the pinning of an individual vortex line on a dislocation core leads to a critical current density jc d (0), the critical current density at higher fields will be equal to j;! d (0) times the fraction of pinned vor- tices n t . The latter is determined by the probability that a given vortex encounters at least one dislocation core in an allowed area Up d jeoa\ determined by equating Up d to the loss of elastic-energy due to the deformation of the vortex lattice,H~tfl so that f c a = f c «(0)n t = 3f(0) exp sd (5) A similar formula was used by Dam et al. to deduce the value of the "characteristic field" B* = &oridUp d / Eq. Due to the non-homogeneous current distribution in su- perconductors, dcpinning from the linear defects would be initiated by the nucleation at the_film surface of vortex kinks joining two dislocation coresO As a result, the ex- perimentally measured low-field "critical" current j sd (0) should be rather smaller than the critical current density in the absence of flux creep and decay rapidly with time. The time decay will be even more rapid when the ap- plied magnetic field and the vortices are not aligned along the dislocation cores. Hence, in analogy to the case of heavy-ion irradiation-induced amorphous tracks, pinning by extended linear screw dislocation cores should lead to a sharp cusp-like maximum in the field orientation- angle dependence of the experimentally measured critical current EMB It should be remarked that both in the case of pin- ning by thickness variations and pinning by screw dis- locations, the low-field critical current flows only at the film surface. Hence, the total screening current as well as 2 the characteristic field of flux penetration H c in a mag- netic experiment (see below) is independent of the film thickness d, and the apparent current density j decreases inversely proportional to d. In what follows, we present a detailed study of the critical current density of YBa2Cu307_a films grown by pulsed laser deposition as function of film thickness, field magnitude and orientation, and temperature. It turns out that the field orientation- and thickness dependence of j c do not provide any evidence for pinning by screw dis- locations or other correlated disorder in our films. Yet, the temperature and field dependence of j c is entirely compar.able.to that measured in films which contain these defectsEM To explain our results, we propose that the rele- vant pinning centers in our films are sparse insulating or normal metallic second-phase inclusions. These are known to exist in sputter-deposited YBa2Cu30j_5 films as platelet- like Y2O3 inclusions of size 15 x 15 nm , with typical densities 10 2 10 2 -3 22 Other authors have reported the presence of Y2O3 inclu- sions in laser-ablated films as well, with typical densities rii 10 22 m~ 3 for inclusions of diameter Di pjr3. — 5 nm, and n, 10 21 m" 3 for Di ~ 10 - 20 nmJ Pin- ning by such large defects turns out to be conveniently described using an extension o£-jthe theory of strong pin- ning of Ovchinnikov and IvlevEj, which we shall develop below ([n]). Our model is, to our knowledge, the only that consistently describes both the field- and thickness de- pendence of the critical current density in YBa2Cu307_,5 films. II. STRONG PINNING BY SPARSE LARGE POINT PINS A. General formulation Strong pinning by large point defects has been de- scribed by simple substitution of the elementary pinning force of a large void f^ max into the collective pinning expression (^).Q This procedure makes the implicit as- sumption that there are many inclusions in a region of volume a\L c . A comparison of the longitudinal correla- tion length, L c ~ 10 nm, which follows from a collective- pinning analysis of experimental critical current density- data, with the expected mean distance between large de- fects, di ~ 30 nm, shows that this approach is inappro- priate. In order to obtain the critical current density for sparse pins, one should not start from the statistical average of the pinning forces of the different defects, but, rather, evaluate the probability that a vortex line will be pinned at all. A nearby defect will be able to trap a vortex line if the gain in pinning energy {J* ps /* max £, is suf- ficient to outweigh the elastic energy loss due to the vortex lattice deformation u(r) caused by displacing the vortex line onto the defect. The maximum allowed lat- eral displacement of the vortex uq determines the "trap- ping area" ~ Uq within which a large defect is an ef- fective pinning site. The bulk (volume) pinning force F p = (ald)~ 1 J2i fp,max i s given by the direct sum of the elementary forces of the individual defects that can effectively pin a single vortex, i.e. those that lie within uq of the vortex lattice position, normalised by the vol- ume a^d available to the vortex. Any vortex line will be pinned on average by N = d/C defects, where C is the av- erage distance between such "effective" pinning centers, so that F, f' f' J p.t aid C a 2 £ (6) The probability to encounter a second effective defect at distance C from a first defect located at z = is given by the product of the probability to encounter none in the interval < z < C and the probability to find at least one at z = C. In the case of layered superconductors considered in Ref. where the only effective defects are those situated in the Cu02 layers, this yields C = s + s [l — cxp (— n n M 2 )] fcexp (— fenoU 2 ,) k=l 1 — exp (— uoUq) (7) and j c = F p /B = C/£ moa ./$o«)[l - exp(-7i n iifj)]. Here tin is the areal density of defects in the Cu02 layers and s is the distance between layers. This result has the same "two-dimensional" form as the result ( |^ ) for linear de- fects extending throughout the thickness of the film. For continuouscl superconductors C = which givesc £exp (— njit 2 £) dC = Jc 1 p. max 2 * — n * u o- $0 (8) (9) In the case of real samples (of finite thickness in the field direction), and especially for thin films, the derivation (||) becomes inappropriate in the limit riiU^d 1, is applicable^rrespective of the defect density and the film thickness. 123 3 B. The trapping area 1. Pin Breaking The trapping area and fp iTnax can be obtained using two different approaches, corresponding to the different mechanisms by which a vortex can be liberated from a defect: pin-breaking and plastic depinningEa The most obvious possibility is "pin-breaking" , in which the ap- plied force must exceed the attractive force fp max ex- erted by a defect on a vortex line. Plastic depinning only occurs if the pins can be considered as "infinitely strong", i.e. fp >max ^> £o, which is not the case in what follows. Exploiting the similarity of pinning by extended pointliloi. defects with pinning by amorphous columnar defects] 2 .] the elementary pinning force of an inclusion of extent (perpendicular to the field direction) Di can be estimated as the product of the fraction of the vortex core volume occupied by the defect and the condensation energy B 2 /2fj, = j£o£~ 2 . Following Ref. ||, the resulting pinning force can be approximated by the interpolation J p , £() In 1 £1 D z 2£ 2 (10) where Df is the extent of the defect along the field di- rection. Note that the above expression corresponds to "<5T c -pinning" p pinning by the variation of the mean-free path in the vicinity of the defect is relatively unimpor- tant, because the quasiparticle scattering probability in the layer of thickness £o surrounding the defect is negli- gible as compared to the total scattering cross-section of the inclusion (£o is the BCS coherence length). The maximum allowed vortex displacement uq is ob- tained by balancing the elastic energy loss with the pin- ning energy gain U p ~ fpmaxC- Taking explicitly into account the range of,the pinning potential , Ovchinnikov and Ivlev obtained!^! 128 ~27 1/4 5/4 5/4 £1/2' (11) Here we have approximated the vortex lattice shear modulus C66 ~ jeodQ 2 and the nonlocal tilt modu- lus C44 w e 2 eoa,Q 2 . Note that for fields in excess of B a = 0.41$ ^ 4 ^ 3 (££o/^p) 10 ^ 3 the maximum allowed vor- tex excursion becomes comparable to ao, and Uq ss a 2 ,. The critical current density follows from combining Eqs. (|), rffij), and (O), 3c « 0.0866ntf e5/4 ^ 1/2 j c « 0.375rHj Df^(T)^ B (B B a ). (16) with B a = Q (eeo/U p ) 2 . For convenience, we shall leave Eqs. (jl|), © and @, ( |l6| ) in their general form, since this permits one to take the possible-effects of thermal smearing of the pinning potentialEirE] into account by simply replacing F(T) by the appropriate temperature dependence. 2. Low fields - single vortex limit A glance at Eqs. — (P~6|) shows that at low fields the trapping area and hence the critical current have an unphysical divergence. The correct low-field limit of the trapping area is obtained by starting from the line tension of a single vortex line e± ~ £ 2 £o- Balancing the energy of elastic deformation of a single line with the pinning energy, S\u 2 / C = U p , one has u 2 = {U p /e\)C and _ _ J Q °° £exp (niU p C 2 /ei) dC J Q exp (niUpC 2 1 £1) dC £1 mU v 1/2 (17) The trapping area becomes Uq = (Up/miEx) 1 / 2 and 1/2 fp,; nUp £ 2 £o 1/2 (18) The single vortex limit is realized for fields such that (Up/nn^ 2 < (U p /ee Q )a , i.e. B < B* = TT^ ni{Up/e ). 4 3. Very thin films A second limit, in which the results of Section II Bl do not hold is that of very thin films of thickness d < Lq, Note that this condition is also violated at very low fields B < $ a s/d 2 ,$ /\ 2 ab . In YBa 2 Cu 3 7 _5 we have Lq(B = 10 mT) « 150 nm, which is comparable to typ- ical film thicknesses. Under these conditions, the proba- bility for a vortex to be trapped by an inclusion is deter- mined by its ability to bend sufficiently within the film thickness, i.e. the tilt contribution dominates the elastic energy. We thus need to repeat our considerations for Lq = d, which will minimize the elastic energy in this case. The total energy of a vortex in the cage of dimen- sions dgd is given by 2 7 ^2 CqqU d + C44—a (19) Equating this to zero yields Uq = d(U p /si), and the crit- ical current density 3c f U p d e 2 e ' (20) Thus, for very thin films, the critical current density should be field-independent and increase linearly with film thickness. Under the condition that B < B* , the crossov er from the "thick thin film" single vortex limit [section ( II B 2| )] and the "thin thin film" limit occurs when the thickness is reduced below the crossover value TTUiUr, 1/2 (21) determined from the equation of the trapping areas Uq in either limit. III. EXPERIMENTAL DETAILS Thin films of YBa2Cu307_5 were deposited on 1.2x1.5 mm 2 LaAlC>3 substrates by KrF laser ablation from a stoichiometric target. The deposition was carried out onto substrates heated to 785° C at 280 mTorr (36.7 Pa) oxygen pressure, in the so-called "off- axis" geometry, i.e. the substrate surface was oriented parallel to the plasma plume axis. From previous experiments, we know this geometry to produce films featuring homogeneous flux penetration (on a fim scale). We have prepared differ- ent series of films of seven different thicknesses, from 100 to 500 nm, using exactly the same deposition procedure. After deposition, the film edge was etched away to pro- duce 0.9x0.9xmm 2 YBa 2 Cu307_(5 squares used for fur- ther investigations. The obtained films were characterized using X-ray diffraction with Ctko radiation and Atomic Force Mi- croscopy (AFM). The diffraction experiments revealed -WOlHU FIG. 1. Tapping mode AFM image showing the surface morphology of a laser-ablated YBa2Cu307_i thin film of 300 nm thickness. the existence (5 atomic % ) of a second (cubic) phase. In AFM experiments we observed a surface morphol- ogy characteristic of island growth, with occasional out- growths (see Fig.Q). The growth islands had a typical lateral size of about 300 nm. This is very similar to the results fouiadrby other authors under the same deposition conditions Ji3li-3 We do not find the spiral growth mecha- nism to be relevant to our samples - a comparison with the deposition conditions of Ref . [l2| indicates that growth spirals are not to be expected. All films were characterized using the magneto-optical technique for flux visualization.^!! A ferrimagnetic garnet film with in-plane anisotropy is placed directly on top of the superconducting film. The garnet is observed using a polarized light microscope with nearly crossed polariz- ers. In this configuration, the reflected light intensity is proportional to the local magnetic induction perpendic- ular to the garnet, allowing the direct observation of flux penetration into the YBa2Cu30y_5 film. The "critical" screening current density j can be obtained directly from the distance Xf between the flux front and the film edge at a given applied magnetic field H a and temperature, j = irH c /d = TrH a /[daxccosh(w/w — Xf)], where w is half the film width and Mr = jd/it the "characteristic field" for flux penetrationE-3 Alternatively, one can fit the po- sition of the flux front as function of the applied field x f 1 cosh (itHa/jd) (22) to obtain the low-field limit of the current density j.El The magneto-optical imaging of the films was used not only to obtain current densities at fields smaller than 300 G, but also to select the most homogeneous films for magnetometry rexperiments using the Local Hall Probe Magnetometer .c3 A commercial miniature GaAs 5 0(1) FIG. 2. Field-orientation dependence of the in-phase fundamental transmittivity for a laser-ablated YBa2Cu307_5 thin film 250 nm thick, for fields of 200 Oe (a) and 2 kOe (b). The temperature is decreased from 86 K down to 79 K in 1 K steps. The ac field had an amplitude of 1.5 Oe and a frequency of 7.75 Hz. Hall probe, placed in the center of the film surface, is used to measure the local induction as function of the applied magnetic field. The result is converted into the so-called "self-field" of the sample, H s = B — H a , cre- ated by the circulating screening current. The latter was determined from the width of the hysteresis loop of H s at constant B and calibrated using the values obtained from the magneto-optic experiments. In this way, the temperature- and field dependences of the screening cur- rent were measured in fields of up to 2 T and tempera- tures between 30 and 85 K. In order to obtain results in the temperature range close to T c , measurements in AC mode were carried out by applying a 1.5 Oe AC field of frequency 7.75 Hz. The AC Hall voltage V(f), measured using a dual-reference lock-in amplifier, was converted to the in-phase fundamental transmittivity of the sample, T' H = [VU,T) - V(f, T -C T c )]/[V(f,T » T c) - V(f,T < T C )]JB Using this technique, the criti- cal temperature of all films was found to be similar, T c ranging between 88.8 and 91.7 K. FIG. 3. Magneto-optical images of flux penetration into laser-ablated YBa2Cu307_^ films of thickness d = 300 nm, 250 nm, 200 nm, 150 nm and 100 nm respectively, at 65 K. Bright areas correspond to a high magnetic induction per- pendicular to the film, while dark regions are areas of low flux density. Horizontal sequences correspond to constant film thickness and varying applied field H a (the values are indicated below each frame), while the vertical sequences are deliberately chosen such that H a /d is approximately constant — a procedure, which reveals the weaker pinning in the thin- ner films. This is most clearly seen for the lower fields. Weak links in the films are revealed as the bright lines of preferential flux penetration. IV. RESULTS Transmittivity experiments as function of DC field ori- entation performed at different temperatures and field magnitudes do not reveal any cusp in the angular depen- dence of the transmittivity when H a is oriented along the film normal (Fig. ||). This indicates that it is un- likely that directional pinning by correlated disorder de- termines the critical current density. If such pinning is due to the dislocations cores, it is masked by another mechanism of strong pinning, which remains to be iden- tified. In order to evaluate the thickness dependence of the critical current density without ambiguities in the con- version of magnetic moment to current density, we turn G 100 c o 10 - 1 1 1 - (a) I I j=4.3xl0 17 d _i i_ 2 10 1 1 10 11 > 8 10 10 \ 6 10 10 4 10 3 10 10 10 3 10 2 10 1 10 50 60 80 100 200 300 400 d ( nm ) 11 :(b) i ' iii i i i i i i i i i i i i i i i i i / • 13 K: / " ~~°25K" / ' -39k; //*^*^\ B 5 1 K" ■ , , , , i , . , , i , . . , i , , . , i , , . . i , , , . ■ 100 200 300 400 500 600 d ( nm ) FIG. 4. (a) Dependence on film thickness d of the characteristic field H c ( • ) and the screening current density j = H c /d ( o ) of laser-ablated YBa2Cu307_,s films in the limit of small flux densities B, at T = 65 K. Data points were obtained from inserting the posi- tion Xf of the flux front at various applied fields in j = nHc/d — 7ri/ a /[darccosh(it;/w — Xf/tv)]. The drawn lines indicate the linear increase j ~ 4.3 x 10 17 d for d < 200 nm, and the plateau reached for larger thicknesses. This behav- ior is in agreement with the predictions (^) and (po[). (b) Film thickness dependence of the screening current density for various temperatures. Data points were obtained by fit- ting the position of the flux front as function of applied field to Eq. (ph to magneto-optical imaging of the flux penetration. Fig. H shows the flux penetration into five films of thick- nesses 100 < d < 300 nm, at T = 65 K. Nearly all films of this batch show at least one weak link, possibly due to the presence of a grain boundary in the substrate. These are easily rendered visible due to the preferential flux penetration along them. Nevertheless, the remain- ing areas feature homogeneous flux penetration and are sufficiently large to obtain reliable results for the flux front position and the screening current density. The im- age frames are chosen so that the vertical sequences have H a /d w constant, a procedure that should yield similar flux penetration if the screening current would be the same for all films. It is clearly seen, however, that for constant H a /d flux penetration is easier in the thinner films. The effect is quantified by measuring the position of the flux-front relative to the film edge and converting 3 10 2.5 10 2 10 1.5 10 o II m 1 10 5 10 o 500 nm • 300 nm □ 250 nm ■ 200 nm O 150 nm ♦ 130 nm T/T c FIG. 5. Compilation of j c (B = 0)-values for 20 films of thickness 130 < d < 500 nm obtained from different growth batches. The data points show the average current density measured on films of the same thickness, while error bars in- dicate the rms deviation in j. this to the characteristic field H c and screening current density j using Eq. (p2[). The result of this procedure is plotted in Fig. [|(a)7^ We find that j manifestly in- creases before reaching a plateau for d > 200 nm. This increase and the subsequent plateau is in contradiction both with the constant j(d) predicted for weak collec- tive pinning (Eq. (Q)), and with the constant H c and j (x g?" 1 expected for pinning by linear defects or by thickness variations (Eq. ( 0)) . However, it is in agree- ment with Eqs. ( |l8| ) and ^2^) derived for sparse inclu- sions or large point pins, which predict that the sus- tainable current should increase with thickness before reaching a plateau for d > d* . The experimental data show a slope 4.3 x 10 17 , which can be compared to the theoretical value ni(fp max /$o)(U p /e 2 £o) predicted by Eq. (gO). This yields, at 65 K, a defect density iii = 1 x Hr 1 m -3 , if we use Eq. ( |Io| ) for the pinning force- Di = 15 nm and Df = 10 nm from TEM experiments,E2l e = 1.7xl0- u (l-t*) Jm-\ £ = 2[(1 + i 2 )/(l - i 2 )] 1 / 2 nm, t = T/T c and T c = 91.7 K. The obtained defect den- sity corresponds to a crossover thickness d* = 60 nm at 65 K, comparable to the experimental data. In order to check the plausibility of the above results, we have prepared two other series of YBa2Cus07_5 thin films using the same nominal deposition conditions. All films in these batches were homogeneous without grain boundaries. Magneto-optical measurements confirm the thickness dependence of the critical current density, as depicted in Figs. 0(b) and [| For small film thicknesses, j increases with d, reaching a plateau for d ~ 200 nm. The current density decreases again for the 500 nm-thick film, perhaps due to the deteriorating epitaxy as one in- creases the film thickness too much. The increase of the critical current density with the film thickness is observed at all temperatures. The only exception are the data for 7 4 10 11 ■ ■ i 1 ■ ■ 1 1 1 1 1 1 1 1 ■ ■ 1 1 ■ ■ 1 1 1 1 1 1 1 1 ■ ■ 1 1 1 (a) • 300 nm 250 nm ■ 200 nm □ 100 nm d = 250 nm 20 30 40 50 60 70 80 90 100 T(K) FIG. 6. (a) Temperature dependence of the low-field sus- tainable current density of laser-ablated YBa2Cu307_5 films of thicknesses 100 nm < d < 300 nm, deduced from fits to Eq. (p2[). The temperature dependence can be very well fit- ted to Eqs. ( |2c| ) (lower curve) and jl^ ) (upper curve - see text). Parameter values are \ a b(0) = 120 nm, £(0) = 2 nm, the defect sizes Di = 15 nm and D\ — 10 nm, B c = 1 T, and a zero-temperature elementary pinning force fp,max{0) = 8x 10 -11 N. (b) the same, on a semi-logarithmic scale, in order to bring out the behavior near T c . the 150-nm thick films, for which j tends to the values measured in thicker films at T < 0.4T C ( open diamonds in Fig. ||). A similar low-temperature behavior can be noted in Fig. 1 of Ref. ||. This may either indicate that a background pinning by other types of defects such as dislocations or oxygen vacancies starts to play a role for T < 35 K, or that the crossover thickness d* decreases with decreasing temperature, a behavior that may be ex- pected from Eqs. jic| ) and (pi]) . The temperature dependence of the low-field criti- cal current density can be rather well described using Eq. @ and the same parameter values as above, as in- dicated by the lower continuous lines in Fig. || (a,b) (for d = 100 nm). Fits using Eq. ( jig ) for larger film thick- nesses are equally successful [upper lines in Fig. raa,b)], although rii = 3 x 10 21 m~ 3 instead of n$ = 1 X 10 m~ 3 should be used. We now turn to the field dependence of the screening current. Typical results are depicted in Fig. ^, 10 11 < 10 10 10' • 30 K o 35 K 40 K A 45 K ■ 50 K □ 55 K A 60 K ♦ 65 K A 70 K o 75 K 0.01 0.1 H H ( T ) FIG. 7. Field dependence of the screening current density of a laser-ablated YBa2Cu307_,s film of thickness 250 nm, at different temperatures 55 < T < 75 K. Continuous lines indicate model fits to j oc B -5 / 8 (Eq jl^)); the power-law prefactor values are plotted in Fig. which shows data for the 250 nm-thick film. These are in every aspect representative of the data for all.^tlipr. films, and very similar to those found previously.Qu3'E-j The sustainable current density shows a low-field plateau of constant current, followed by a smooth decrease of j for fields higher than a given threshold. This decrease was re- cently interpreted in terms of creep of unpinned vortices through the forest of vortices presumably trapped on screw dislocation cores, a mechanism that ought to lead to a ^-dependence of the screening current density. However, the same field dependence is easily explained in terms of strong pinning, see Eqs. ( |i"2| ) and (|l5|). The straight lines in Fig. |?] show fits to the Z? -5 / 8 -dependence predicted by Ovchinnikov and Ivlev; the power-law pref- actors are given in Fig. ^. Adapting the same parameter values as for the fits to the zero-field current density, one sees that the temperature dependence of the prefac- tor withstands the comparison with the prediction jl2] ) rather well. Again using the same parameters for the YBa 2 Cu307_5 material and the size of the Y2O3 inclu- sions, we obtain the drawn line of Fig. || for rii = 2.6 xlO 21 m~ 3 , very close to the previously obtained values. While the temperature dependence predicted by Eq. (|l^) well matches the data at higher temperatures, it fails at tem- peratures below ~ 35 K, again indicating that pinning by other mechanisms may become important there. Note that the field dependence of the sustainable cur- rent density does not follow a pure power law by any means, but gradually bends over from a _B~ 5 / 8 -behavior to a 1/B dependence. This dependence is again in agree- ment with the strong pinning scenario: at high fields the vortex displacements due to pinning become comparable to the vortex spacing and the critical current density 8 13 3 10 2.5 10 2 10 £ 1.5 10 1J - 1 10 5 10 I I I I I I • 300 nm o 250 nm ■ 200 nm □ 100 nm 2.4xl0 21 m 3 ' 100 ,1,-1 R 5/8 50 60 70 T(K) FIG. 8. Prefactors rnulf^ ni} l v ^ax^o 1 ( c/ p/ e£ o) 5/4 ($o /8 /C 1/2 ) obtained from the fits in Fig. 0. the drawn line indicates a fit to Eq. ( |l2] ) using the same parameter values as in Fig. [] (A a (,(0) = 120 nm, £(0) = 2 nm, B c = 1 T, Di = 15 nm, D% = 10 nm) and m = 2.6 x 10 21 starts do decrease more rapidly, as j c oc B^ 1 (Eq. ([L3|)). The magnitude of the field B a at which this effect is ex- pected is typically of the order 1 T. V. DISCUSSION The above results show that, even if perfect agreement is not found, strong pinning by Y2O3 inclusions can very well explain the large magnitude, as well as the tem- perature, field, and thickness dependence of the critical current density in laser ablated YE^CusOt-^ films. The primary indication for this is the observed dependence on film thickness of the screening current density: j first in- creases as function of film thickness before saturating at d w 250 nm, and eventually slowly falling off again. This dependence is very similar to that measured by Jooss et al.t2 Other authors also report the increase of j as func- tion of film thickness d, but with a maximu m at fLsi 100 nm followed by a munh-more rapid drop of j l 15 H 3 ^-P 7 l Pro- posed explanationsEEO have been the increase of the ef- fective penetration depth from X a b towards 2\ 2 ab /d for films thinner than 2A a b,E3 a bad film morphology and ill- connected islands for d < 100 nm£3 and differences in dp. feet structure between the thinner and the thicker films.EZI We note that the fact that films from different sources display maximum j for widely different film thicknesses, 100 nm and 250 nm respectively, exclude that the relative magnitude of d and X a b is at the origin of the increase of j(d). However, our model explains this difference quite naturally in terms of Eq. (|l]) and the different density of second-phase inclusions in different films. An effective penetration depth 2\ 2 b /d would modify the current den- sity obtained from pinning by thickness variations, but not the pinning force in the case of core pinning, for in- stance, by extended point defects or linear defects. The differences in optimal thickness also seem to exclude bad film morphology as an explanation: while such an effect may be likely for very thin films of thickness much less than 100 nm, it cannot explain an increase of j(d) up to d — 200 nm. Finally, the role of deep trenches that mod- ify the total current circulating in thin filmsEZI is excluded by the magneto-optical observations, which show homo- geneous, featureless flux penetration. Thus, we conclude that the observed thickness dependence is in agreement with the scenario of strong pinning by extended point defects only. Further evidence is the the lack of a cusp-like angu- lar dependence of the critical current density for field aligned close to the film normal, which argues against a predominant role of extended line-like defects such as screw dislocation cores. The large difference between the algebraic temperature dependence of j c with the ex- ponential dependence observed in clean YBa2Cua07_5 single crystals renders the relevance of weak collective pinning by oxygen vacancies unlikely. These observa- tions gain in importance if we note that the field and temperature dependences reported here are observed in YBa 2 Cu307_(5 films quite irrespective of the deposition method and conditions. While generally neglected, small Y2O3 inclusions are ubiquitous in YBa2Cu307_,5 films deposited by different methods and may provide an ex- planation for both the magnitude and behavior of j c and the generality of this behavior. In our case, the defect density needed to explain the experimentally observed j c , 1 x 10 21 < tii < 3 x 10 21 m~ 3 means that the pre- cipitates occupy between 0.2% and 0.6% of the sample volume. The 5% of secondary phase material found in the X-ray analysis would be present as larger CuC>2 out- growths which do not pin vortices efficiently. Strong pinning by larger point-like inclusions domi- nates over collective pinning by oxygen vacancies because thermal fluctuations in high temperature superconduc- tors efficiently smear out the pinning potential of the vacancies£3 Even at zero temperature it is energetically more favorable to pin a vortex on rather large inclusions than to have it lower its energy by wandering through the weak collective pinning landscape. Restricting ourselves to the single vortex limit, pinning on a large inclusion yields an energy gain jEqD^; this should be compared to the potential energy gain due to pinning by oxygen vacancies in the volume UqC Strong pinning by Y2O3 inclusions is more favorable if 2C-1 (23) where U v is the pinning energy gained in a correlation vol- ume due to weak collective pinning by the oxggen vacan- cies and C ~ 0.63 is the wandering exponent.B Substitut- ing C from Eq. (|l7|), assuming U p = ea(jQe 2 n v D^ 2 ) 1 ^ 3 for <5K-pinning (pinning by mean free path fluctuations), and the oxygen ion radius D v sa 3 x 10~ 10 m, we find that for physically allowable oxygen vacancy densities n v strong pinning by Y2O3 inclusions is always more favor- able. In the above, we have disregarded the effects of flux creep. While this can more or less be justified at low temperatures where creep rates are small, at higher tem- peratures flux creep certainly determines the observed sustainable current density. Experiments by Klaassen et aZ.tj show that in the low-field regime of constant cur- rent density the activation energy for flux creep E a is of the order of 600 K; the activation barrier rapidly de- creases and becomes constant at fields where j decreases as function of B. This low value of E a , much lower than the value predicted by nucleation-type ■ cree p models for depinning from linear extended defect s,E£iEd could be ex- plained by the fact that vortices are held not by extended defects but by second-phase inclusions of much shorter longitudinal dimensions. The rapid decrease of the ac- tivation energy with magnetic field would be due to the decrease in the average number of defects, by which any given vortex line is effectively pinned (i.e. C becomes of the order of the film thickness). The high-field regime of constant activation barrier can be explained if one as- sumes that there is less than one effective pin per vortex line; Our experimental result rii ss 1 x 10 21 m~ 3 implies a distance between defects of the order 100 nm, compara- ble to both the smallest film thicknesses and the vortex spacing at B = 0.2 T. So, indeed, in the field regime of decreasing j(B), there is less than one effective defect per vortex, and the flux creep activation energy should be given by the energy barrier needed to break free a line from one defect. The power-law field dependence implies that the fraction of vortices pinned in this way suffices to keep the whole vortex lattice at rest; at the threshold (critical) current density, the entire vortex lattice depins until a vortex line is trapped by the next defect. Only at higher fields, where the experimental j(-B)-data sig- nificantly deviate from a power-law, do we expect creep effects, vortex lattice shear, and plastic deformations of the vortex lattice to become important. VI. CONCLUSIONS We have measured the dependence of the sustainable current density in pulscd-laser deposited YBa2Cu307_,5 thin films on film thickness, field orientation and magni- tude, and temperature. While the field and temperature dependence of the current density show the qualitative behavior often reported in literature for laser ablated, sputtered, and MO-CVD thin films, the film thickness dependence is consistent only with strong pinning by sparse large second phase inclusions (such as Y2O3). A model description of such pinning in different limits of flux line density and film thickness is consistent with the experimentally found behavior, if one assumes the inclu- sions to have the typical sizeJL5_j< 15 x 10 nm 3 obtained from previous TEM studiesJ^SEj, with a defect density rii ~ 1 — 3 x 10 21 m~ 3 . Such a defect density suggests in turn that at high fields, only a fraction of the vortices are pinned; these suffice to hold the entire lattice at rest until the critical current density is reached. VII. ACKNOWLEDGEMENTS This work was partially supported by Polish Govern- ment grant No." PBZ-KBN/013/T08/19. Support from French - Polish Programme "Polonium" and ESF "VOR- TEX programme is also gratefully acknowledged. B. Roas, L. Schultz, and G. Saemann-Ischenko , Phys. Rev. Lett. 64, 479 (1990). 2 P.H. Kes, in Phenomenology and Applications of High Tem- perature Superconductors, Ed. K. Bedell et al., (Addison- Wesley, New York 1992). 3 R. Griessen, Wen Hai-hu, A.J.J, van Dalen, B. Dam, J. Rector, HG. Schnack, S. Libbrecht, E. Osquiguil, and Y. Bruynseraede, Phys. Rev. Lett. 72, 1910 (1994). 4 A.I. Larkin and Yu. Ovchinnikov, Journ. Low Temp. Physics (1979) 5 G. Blatter, M.V. Feigel'man, V.B. Geshkenbein, A.I. Larkin, and V.M.Vinokur, Rev. Mod. Phys. 66, 1125 (1994). 6 H.R.A. Douwes, Ph.D. thesis, Leiden University, the Netherlands (1994). 7 P.H. Kes and C.C. Tsuei, Phys. Rev. Lett. 47, 1930 (1981). ibid. Phys Rev. B 28, 5126 (1983). 8 S. Kokkaliaris, A. A. Zhukov, P.A.J, de Groot, R. Gagnon, L. Taillefer, and T. Wolf, Phys. Rev. B 61, 3655 (2000). 9 M. Hawley, I.D. Raistrick, J.C. Beery, and R.J. Houlton, Science 251, 1587 (1991). Ch. Gerber, D. Anselmetti, J.G Bednorz, J. Mannhart, and DG. Schlom, Nature 350, 280 (1991). 1 I.D. Raistrick and M. Hawley, in Interfaces in High T c Su- perconductors, Springer (1993). 2 B. Dam et al, Physica C 296, 179 (1998); B. Dam, J.H. Rector, J.M. Huijbregtse, and R. Griessen, Physica C 305, 1 (1998). 3 Ch. Jooss, A. Forkl, R. Warthmann, H.-U. Habermeier, B. Leibold, and H. Kronrmiller, Physica (Amsterdam) C 266, 235 (1996). 4 B. Dam, J.M. Huijbregtse, F.C. Klaassen, R.C.F. van der Geest, G. Doornbos, J.H. Rector, A.M. Testa, S. Freisem, J.C. Martinez, B.Stauble-Pumpin, and R. Griessen, Nature 399, 439 (1999). 5 F.C. Klaassen, Ph.D. thesis, Vrije Universiteit, Amsterdam (2002). F. C. Klaassen, G. Doornbos, J. M. Huijbregtse, R. C. F. van der Geest, B. Dam, and R. Griessen, Phys. Rev. B 64, 184523 (2001). 10 16 A. Wahl, V. Hardy, J. Provost, C. Simon, and A. Buzdin, Physica C 250, 163 (1995). 17 V. Hardy, S. Hebert, M. Hervieu, Ch. Simon, J. Provost, A. Wahl, and A. Ruyter, Phys. Rev. B 58, 15218 (1998). 18 C.J. van der Beek, M. Konczykowski, R.J. Drost, P.H. Kes, N. Chikumoto, and S. Bouffard, Phys. Rev. B 61, 4259 (2000). 19 R. Prozorov, E.B. Sonin, E. Sheriff, A. Shaulov, and Y. Yeshurun, Phys. Rev. B 57, 13845 (1998). 20 D.R. Nelson and V.M. Vinokur, Phys. Rev. Lett. 68, 2398 (1992) . 21 D.R. Nelson and V.M. Vinokur, Phys. Rev. B 48, 13060 (1993) . 22 T.I. Selinder, U. Helmersson, Z. Han, J.-E. Sundgren, H. Sjostrom, and L.R. Wallenberg, Physica (Amsterdam) C 202, 69 (1992). 23 A. Catana, R.F. Broom, J.G. Bednorz, J. Mannhart, and D.G. Schlom, Appl. Phys. Lett. 60, 1016 (1992). 24 G. Kastner, D. Hesse, R. Scholz, H. Koch, F. Ludwig, M. Lorenz, and H. Kittel, Physica (Amsterdam) C 243, 281 (1995). 25 K. Verbist, A. Kiihle, and A. Vassiliev, Physica (Amster- dam) C 269, 131 (1996). 26 Yu. N. Ovchinnikov and B.I. Ivlev, Phys. Rev. B 43, 8024 (1991). 27 In fact, evaluating the probability V>i = 1 — exp(— mu^d) « niU%d that a vortex line will encounter at least one defect overestimates the resulting critical cur- rent density j c — P>ifp, ma x/&o- The real critical cur- rent lies between this and the product Vifp imax /$o, where Pi = cxp(— niU^d) is the probability to encounter one pin- ning center only. This may explain why the experimental j c deviates downwards from a power-law behavior at high fields. An (incorrect) formula for j c that attempts to take this into account can be found in I. Abalosheva et al, Pro- ceedings of the Second Euroconference on Vortex Matter in Superconductors , Physica (Amsterdam) C 369, 240 (2002). 28 A. Schonenberger, A. Larkin, E. Heeb, V. Geshkenbein, and G. Blatter, Phys. Rev. Lett. 77, 4636 (1996). 29 V.M. Vinokur, B. Khaykovich, E. Zeldov, M. Kon- czykowski, R.A. Doyle, and P.H. Kes, Physica C 295, 209 (1998). 30 M.V. Feigel'man and V.M. Vinokur, Phys. Rev. B 41, 8986 (1990). 31 L.A. Dorosinskii, M.V.Indenbom, V.I. Nikitenko, Yu.A. Ossip'yan, A. A. Polyanskii, and V.K. Vlasko-Vlasov, Phys- ica (Amsterdam) C 203, 149 (1992). 32 M. Konczykowski, F. Holtzberg, and P. Lejay, Superc. Sci. Tech. 4, S331 (1991). 33 J. Gilchrist and M. Konczykowski, Physica (Amsterdam) C 212, 43 (1993). 34 E.H. Brandt, M.V. Indenbom and A. Forkl, Europhys.Lett. 22, 735 (1993). 35 M.V. Feigel'man and V.M. Vinokur, Phys. Rev. B 41, 8986 (1990). 36 C.J. Liu, C. Schlenker, J. Schubert, and B. Stritzker, Phys. Rev. B 48, 13911 (1993). 37 E. Sheriff, R. Prozorov, Y. Yeshurun, A. Shaulov, G. Ko- ren, and C. Chabaud-Villard, J. Appl. Phys. 82, 4417 (1997). 38 J. Pearl, Appl. Phys. Lett. 5, 65 (1964). 39 F.C. Klaassen, G. Doornbos, J.M. Huijbregtse, B. Dam, and R. Griessen, Physica (Amsterdam) C 341-348, 1463 (2000). 11